Enhancing Phishing Detection: A Multi-Layer Ensemble Approach Integrating Machine Learning for Robust Cybersecurity
Candra Ahmadi, Jiann-Liang Chen · 2024
Cybersecurity is a critical concern in our increasingly digital world, with phishing attacks posing one of the most insidious threats to individual and organizational security. Although machine learning has revolutionized phishing detection, there is still a considerable gap in the ability to detect such threats in real time with high accuracy and efficiency. Current methods often fail to dynamically adapt to the rapidly evolving tactics of cyber adversaries, leading to gaps in detection and prevention capabilities. We introduce an optimized Multi-Layer Ensemble Model that leverages a combination of advanced machine learning classifiers, including K-nearest neighbors, Decision Trees, Random Forest, Extra Tree, and XGBoost, to enhance the accuracy and efficiency of phishing website detection. Experimental results demonstrate that our model achieves a significant improvement in detection accuracy, rising from 95.99% to 97.25%, while maintaining a minimal response time of 4.8 seconds. This model effectively reduces both false positives and negatives, adapting dynamically to new phishing tactics. This advancement not only addresses critical vulnerabilities in cybersecurity defenses but also sets a new benchmark for real-time phishing detection systems, offering a scalable solution that can evolve with the threat landscape.